一种新的三维纯角度无偏定位跟踪算法

三维空间中多站纯角度无源定位跟踪实质上是一个非线性状态估计问题。经典的扩展卡尔曼滤波(EKF)算法等递推类算法性能不稳定,而将非线性观测方程转化为伪线性方程会产生有偏估计。为了实现快速稳定无偏定位跟踪,提出了一种新的双站纯角度快速渐进无偏定位跟踪算法。该算法通过对未知状态变量含二次约束的伪线性方程进行约束最小二乘(CLS)极小化处理,最终只需要对一对矩阵束进行广义特征分解即可获得目标状态估计值。仿真结果表明,该算法与EKF算法及最小二乘(LS)算法相比,定位性能更稳定,精度更高,在测量误差较大或者两个观测站测量误差不一致时优势更明显。

A novel asymptotically unbiased algorithm for passive location and tracking with angles-only measurements

The multi-station angles-only passive location and tracking in 3-D space is a nonlinear state estimation problem in essence. The classic extended Kalman filtering （EKF） algorithm is unstable and the pseudo linear algorithm derived from measure equations will result in biased or divergent solution. A new asymptotically unbiased location and tracking algorithm with angles-only measurements is proposed to solve these problems. The proposed algorithm uses constraint least squares minimization on the pseudo linear equation which includes quadratic constraint about the state vector and the state estimate can be got by taking a generalized eigen-decomposition to a pair of matrix pencil.